-
Notifications
You must be signed in to change notification settings - Fork 3
Options Structure
Paul Wright edited this page Aug 10, 2021
·
13 revisions
The RunPreproc
function takes two arguments: paths
, a cell array that specifies the input image filenames, and opt
, a structure that specifies the options. The fields of the opt
structure are detailed here. The subfields of do
are all false
by default.
Option | Type | Description | Notes |
---|---|---|---|
opt.dir_out |
string | Output directory | Default: 'output' . Files in ‘path’ are copied to the output directory. Files ending ‘.gz’ are gunzipped (in which case outputs will be gzipped). Regular images are written int16. If label images are specified, these are checked so that values are <= 255 and written uint8 . |
opt.prefix |
string | Filename prefix | Default: 'pp_' . Added to filenames during initial copy. |
opt.labels.part |
multi | Values to use to collapse labels. | First, rounds label image values. Next, if part is a string, binarises label values to zero and nonzero. If part is a vector, assigns sequential values to voxels with each value in the vector. If part is a cell array, assigns sequential labels to voxels with any of the values in each cell. |
Option | Type | Description | Notes |
---|---|---|---|
opt.do.erode |
Boolean | Enable erode step | Removes a few of the outer voxels |
Option | Type | Description | Notes |
---|---|---|---|
opt.do.res_orig |
Boolean | Enable resetting origin | Important for CT |
opt.reset_orig.only_neg |
Boolean | Only reset origin if image contains negative values. | Negative values mean the image might be a CT. |
Option | Type | Description | Notes |
---|---|---|---|
opt.do.real_mni |
Boolean | Enable realignment | |
opt.realign2mni.rigid |
Boolean | Make realignment rigid | Default: true . |
Option | Type | Description | Notes |
---|---|---|---|
opt.do.nm_reorient |
Boolean | Enable reslicing | The orientation matrix is applied to the image and reset. |
Option | Type | Description | Notes |
---|---|---|---|
opt.do.crop |
Boolean | Enable cropping | Remove unneccesary data to reduce image size |
opt.crop.keep_neck |
Boolean | Keep neck when cropping | Default: true . False makes for even smaller image |
Option | Type | Description | Notes |
---|---|---|---|
opt.do.coreg |
Boolean | Enable coregistration | Uses Normalised Mutual Information (NMI) |
opt.coreg.ref |
integer | Reference image | Default: 1 . Which of the images acts as target for the others? Specify index in cell array paths{1} . |
Option | Type | Description | Notes |
---|---|---|---|
opt.do.denoise |
Boolean | Enable denoising |
Option | Type | Description | Notes |
---|---|---|---|
opt.do.superres |
Boolean | Enable super-resolution | |
opt.superres.ix |
integer vector (N img) | Index of images to super-resolve | Vector the same length as cell array of images. Number indicates channel (i.e. modality). Examples: [1 1 1] specifies three observations of one channel. [1 1 2 2 3 3] specifies two observations each of three channels. Default = [] , which is interpreted as one observation from each channel. |
opt.superres.verbose |
Boolean | Enable verbose output from super-resolution | Default: false . |
Option | Type | Description | Notes |
---|---|---|---|
opt.do.reslice |
Boolean | Reslice images to have same image grids | Only done if super-resolution is disabled. |
opt.reslice.ref |
string | Reference image | Default: 1 . Which of the images acts as target for the others? Specify index in cell array paths{1} if labels specified. |
opt.reslice.deg |
integer | Interpolation order | Default: 1 . 0 = nearest neighbour. 1 = trilinear. >=2 gives degree of B-spline. |
Option | Type | Description | Notes |
---|---|---|---|
opt.do.vx |
Boolean | Change voxel size | If super-resolution is disabled, change voxel size when resetting origin, applying bounding box, or in a separate step (see detailed pipeline). |
opt.vx.size |
number | Specify voxel size in mm. | Default: 1 . |
opt.vx.deg |
integer | Interpolation order | Default: 1 . 0 = nearest neighbour. 1 = trilinear. >=2 gives degree of B-spline. |
Option | Type | Description | Notes |
---|---|---|---|
opt.do.bb_spm |
Boolean | Make image have same bounding-box as default SPM12 template | Requires realignment to MNI. |
opt.bb.dim |
[integer,integer,integer] | Output dimensions after applying bounding-box | By default the atlas dimensions will be used |
Option | Type | Description | Notes |
---|---|---|---|
opt.do.segment |
Boolean | Enable unified segmentation | |
opt.segment.write_tc |
Boolean matrix (6x4) | Select tissue classes to write | Default native space GM, WM and CSF. Rows: tissue classes. Columns: image to write (native space, DARTEL import, standard space unmodulated, standard space modulated). |
opt.segment.write_bf |
Boolean | Write bias correction results | Default: false . Can be a scalar, column vector length 2, or matrix N images x 2. Columns: write bias corrected image, write bias field. |
opt.segment.write_df |
Boolean | Write deformation fields | Default: false . Can be scalar or column vector length 2. 1: write forward. 2: write inverse. |
opt.segment.dir_out |
string | Directory to save segmentation output | |
opt.segment.make_4d |
Boolean | Write 4D nifti with segmentations. | |
opt.segment.domask |
Boolean | Enable cost function masking | Typically, a lesion mask is used to exclude abnormal voxels from segmentation. |
opt.segment.maskvol |
integer | Mask volume | Specify index in cell array paths . |
Option | Type | Description | Notes |
---|---|---|---|
opt.do.bfcorr |
Boolean | Enable bias field correction | Depends on segmentation |
Option | Type | Description | Notes |
---|---|---|---|
opt.do.skullstrip |
Boolean | Skull strip | Depends on segmentation |
Option | Type | Description | Notes |
---|---|---|---|
opt.pth_template |
string | Path to template | Used to reslice and affinely register images to a template. |
Option | Type | Description | Notes |
---|---|---|---|
opt.do.normalise |
Boolean | Enable writing normalised images | |
opt.normalise.vox |
integer or column vector (3) | Specify voxel size | If integer, outputs cubic voxels this size. If row vector uses these dimensions. Default from SPM (usually [2 2 2] ). |
opt.normalise.deg |
integer | Interpolation order | Default: spm_get_defaults('normalise.write.interp'). 0 = nearest neighbour. 1 = trilinear. >=2 gives degree of B-spline. |
Option | Type | Description | Notes |
---|---|---|---|
opt.do.int_norm |
Boolean | Enable intensity normalisation | |
opt.int_norm.rng |
vector (2) | Range into which to normalise intensities | Default: [0 1]
|
Option | Type | Description | Notes |
---|---|---|---|
opt.do.write2d |
Boolean | Write 2D versions | If segmentation was run, writes 2D versions of segmented images too. |
opt.dir_out2d |
string | Directory to save 2D versions | |
opt.write2d.deg |
integer | Interpolation order | Default: 0 . 0 = nearest neighbour. 1 = trilinear. >=2 gives degree of B-spline. |
opt.write2d.sliceix |
integer | Slice index | This slice is written to the 2D file for each axis. |
Option | Type | Description | Notes |
---|---|---|---|
opt.do.go2native |
Boolean | Write images in native space | |
opt.do.writemat |
Boolean | Save orientation matrix to .mat file |